The aim of the study was to develop a methodology to identify the best use of a longitudinally measured biomarker in relevance to prognosis.
Materials and Methods
Data of squamous cell carcinoma
) from 770 patients with cervical squamous cell carcinoma
(SCC) were used. The pretreatment, nadir, and time-dependent SCC-Ag
values were analyzed in relevance to disease relapse and death with univariate and multivariate analysis side by side with a variety of available clinicopathologic factors. The predictive power of the significant variates was evaluated by C-index with 5-fold cross validation.
The pretreatment, nadir, and time-dependent SCC-Ag
were all significant risk factors for both relapse and death in the univariate analysis (p
< .05), and the time-dependent SCC-Ag
had the highest C-index in both events. The nadir and time-dependent SCC-Ag
were both independently significant in response to relapse with International Federation of Gynecology and Obstetrics (FIGO) stage as the covariate, and the latter had a higher C-index (0.745). Only the time-dependent SCC-Ag
was independently significant together with FIGO stage in response to death with the C-index at 0.844.
Increases in the serum level of SCC-Ag
in cervical SCC patients suggest a higher risk of both relapse and death. The best use of serial SCC-Ag
measurements is to include the time-dependent value in prognostic assessment with FIGO stage also accounted for. Cervical SCC patients should be followed up on their levels of SCC-Ag
, and prognostic evaluation should be updated with recent measurements.